LEDs have many potential applications. For example, LEDs can be used in displays, such as LCDs and in projection, entertainment, general lighting and automotive applications where high brightness and compact illumination is required. The benefits of LEDs over conventional incandescent lighting and halogen lighting are high brightness, long life, instant operation, energy saving, environmental friendliness, durability and compactness.
Notwithstanding those benefits, most of the light generated inside a conventional LED cannot be efficiently extracted from the active layer. Almost 80% of the light generated in an LED is outside the escape cone of the structure. Most of the light remains either guided in the core or totally internally reflected in the high refractive index substrate layer.
The potential for increasing the efficiency, and in particular the light extraction efficiency, of LEDs has long been recognised. For example, the difference between the refractive index of a high index substrate (n˜3.5) and that of the epoxy used to encapsulate the LED (n˜1.5) is large, resulting in a relatively small critical angle for total internal reflection. This in turn dramatically restricts the external quantum efficiency compared to that of the internal quantum efficiency. It has been realised that using an optically transparent conductive layer and a cladding with a low refractive index and transmitting substrates with a low refractive index improves light extraction efficiencies.
FIG. 1 illustrates a method that has been used to improve directionality of light emitted from the LED structure. FIG. 1a shows the 6 light escape cones from the active layer of an LED. FIG. 1b shows the use of a Distributed Bragg Reflector 104 located underneath the active layer, which reflects light back up and out of the LED. The use of a Distributed Bragg Reflector is used to direct more light out of the top of the LED structure emitting power in a specific angular cone. This technique is described in U.S. Pat. No. 6,015,719.
Alternatively, or additionally, the use of microlens arrays placed on the top surface of an LED structure can provide enhanced extraction. This was first proposed in U.S. Pat. No. 5,087,949. It has also been suggested by S Moller et al. in Journal of Applied Physics 91, 3324 that attaching the microlens array on an organic LED (OLED) glass substrate can provide similar benefits to those found with semiconductors LEDs, where an external coupling efficiency improvement of x2.3 across the complete viewing half space was observed.
An illustration of the use of microlens arrays on the surface of an LED structure is shown in FIG. 2. Numeral 201 shows the out of plane coupling of light. A microlens array 202 is placed on the top of a glass substrate 203 which covers the active layer of a conventional LED.
FIG. 3 shows the use of angled facets to preferentially reflect light combined in the active layer out of the top surface of the structure. The active layer 302 and the overlying cladding layer have tapered side walls 305. A metal contact 303 is shown on the top surface. The light generated in the active layer reflects off the walls 305 and exits the top of the structure. This is described in U.S. Pat. No. 6,015,719.
The use of high index polymers that are optically clear can significantly reduce reflection losses at the semiconductor substrate/air interface. This is illustrated in FIG. 4. FIG. 4a shows a conventional low index gel 401 on an LED. The escape cone angle from the active layer 404 is shown as 402. The fundamental waveguide mode angle 403 lies outside the escape cone. FIG. 4b shows the use of a high index gel. The use of a high index gel 406 provides a reduced refractive index contrast and hence provides a larger escape cone angle 407 for the totally internally reflected light. The fundamental waveguide mode angle 405 now lies within the escape cone. A light output increase of around 20% can be achieved by the variation of refractive index from 1.46 to 1.60.
Another approach taken to improve the extraction efficiency of LEDs is taught by Schnitzer et al in Applied Physics Letters 63, 2174 (1993). This paper describes the use of random texturing or roughening of the surface of the semiconductor LED as shown in FIG. 5. Referring to FIG. 5, the roughening 503 on the surface of the LED provides multiple miniature domains with different escape angles. When the totally internally reflected light from the active layer 502 is incident on one of those surfaces it has an increased probability of lying in the escape angle of that surface as compared to a totally flat surface. This provides an improved extraction efficiency. While this method is efficient at extracting light that is experiencing multiple total internal reflections due to absorptive regions within an LED, the light is rapidly attenuated and hence does not contribute to orders of magnitude light extraction and improvement.
In U.S. Pat. No. 5,779,924 the use of periodic texturing on at least one interface of the structure is described and is suggested to improve the extraction of light out of the active core layer 603. This is shown in FIG. 6. The periodic texturing 602 directs more light out of the structure without totally internally reflecting the light inside the structure, where it is greatly attenuated.
Instead of periodic texturing, photonic crystals have been used to achieve the same effect of enhanced light extraction. This is described in U.S. Pat. No. 5,955,749.
Surface roughening, periodic texturing and regular photonic crystals all enhance light extraction from LEDs through the same mechanism, that of modifying the surface profile to improve the probability that light generated in the active layer incident on the surface will be incident at an angle to the surface which allows it to escape from the structure.
Regular photonic crystals (PCs) can also lead to greater light extraction via another mechanism. It is well known that it is generally not desirable to etch into the active layer due to increased surface recombination of carriers, which affects the overall photoluminescence quantum efficiency of the active layer. Nevertheless, if the PC is in close proximity to the active layer it is possible to enhance the rate of spontaneous emission through the Purcell effect. In the Purcell effect it is suggested that the spontaneous emission of an atom placed within a wavelength-sized microcavity can be increased when compared to a bulk structure. A regular photonic crystal 704, as shown in FIG. 7, can confine an optical mode in a “cavity” 707 by virtue of the effective refractive index contrast experienced by the optical mode in the active layer 705. The emitted light is shown as 703.
Erchak et al, in App. Phys. Lett. Vol. 78, no. 5, 29 Jan. 2001, Pg. 563-565 have reported a 6-fold increase in light due to increased extraction and radiation efficiency by the use of a PC embedded in an LED.
Alternative proposed designs suggest grating type PC structures. In this set-up it is suggested that the direct transmitted mode drains only 20% of the total photons while 50% of the light is confined in high index guided modes. The high-index guided modes are very efficiently coupled into external modes and launched out of the structure.
However, due to the highly diffractive nature of regular photonic crystals, the far field emission out of the top of LED structures is localised into Bragg spots (which follows the periodic lattice nature of the PC structure), as shown in FIG. 8. FIG. 8 shows in cross-section a regular photonic crystal 802 in an LED structure with an active layer 804, as shown in FIG. 7. The structure is also shown from above. The out of plane coupled light 801 forms a far filed illumination pattern 805 consisting of a number of Bragg spots. For most applications it is desirable to achieve more even illumination in the far field.
It is an object of the present invention to provide improved light extraction from LEDs whilst obtaining desirable far-field illumination.